Effects of green space on physical activity and body weight status among Chinese adults: a systematic review

Background Green space may provide many benefits to residents’ health behaviors and body weight status, but the evidence is still relatively scattered among Chinese adults. The purpose of this study was to review the scientific evidence on the effects of green space on physical activity (PA) and body weight status among Chinese adults. Methods A keyword and reference search was conducted in Pubmed, Web of Science, MEDLINE, and PsycINFO. Studies examining the associations between green space and PA, body mass index (BMI) among Chinese adults were included. The quality of the included literature was evaluated using the National Institutes of Health’s Observational Cohort and Cross-Sectional Study Quality Assessment Tool. Results A total of 31 studies were included that met the inclusion criteria, including 25 studies with a cross-sectional design, 3 studies with a longitudinal design, and 3 studies with an experimental design. Street-level green view index and green space accessibility were found to be positively associated with PA, but negatively associated with BMI. In most studies, there was a correlation between green space ratio in local areas and BMI. In addition, green space interventions were effective in increasing PA and decreasing BMI among Chinese adults. In contrast, further evidence is needed to support the association between the design characteristics of green space and PA and BMI. Conclusion Preliminary evidence suggests that green space has a positive effect on PA and BMI among Chinese adults. However, there are contradictory findings, and future studies adopting longitudinal and quasi-experimental studies are needed to further explore the causal relationship between green space and PA and BMI to provide a relevant theoretical basis for policymakers.


Introduction
There is evidence that physical inactivity is associated with an increased risk of chronic disease, leading to obesity, diabetes, hypertension, heart disease, and other chronic diseases (1). Conversely, sufficient physical activity (PA) and regular PA can prevent and reduce the risk of chronic disease and contribute to a higher quality of life. PA is defined as physical movement in which muscle contraction causes energy expenditure (2). The energy expenditure of a person while sitting still is approximately 1 metabolic equivalent (MET). Moderate physical activity (MPA) usually has a range of ≥3METs and < 6METs (e.g., brisk walking, dancing, etc.); moderate to vigorous physical activity (MVPA) is defined as requiring ≥3METs of energy expenditure (e.g., brisk running, fast cycling, etc.) (3). The World Health Organization (WHO) reports that physical inactivity is the fourth leading risk factor for global mortality, further contributing to the prevalence of non-communicable diseases (4). In addition, previous studies show that nearly 2 billion adults worldwide are considered overweight, more than half of whom are classified as obese (5,6). Obesity is associated with the development of many diseases such as atherosclerosis, type 2 diabetes, and dyslipidemia, and has become a global epidemic and public health crisis (7,8). Previous studies have reported that chronic diseases caused by physical inactivity and obesity not only have a serious negative impact on the physical and mental health of individuals but also create a heavy burden on social development (9)(10)(11)(12). Therefore, WHO and professional health agencies expect to develop active and effective intervention policies to help people improve their PA and adopt healthy weight.
Available evidence suggests that environmental interventions can help people achieve a balanced lifestyle in their daily lives (13)(14)(15). In terms of PA, it can be done in an indoor environment or an outdoor environment. However, existing research has found that PA in an outdoor environment has greater benefits for human physical and mental health than in an indoor environment (16). Green spaces are an important part of the outdoor environment and include greenways, parks, gardens, street greenery and land covered with grass, shrubs, trees, or other vegetation (17). Green spaces are associated with PA among residents (18,19). Several studies have shown that green spaces have a positive impact on people's physical and mental health, and can promote PA, reduce sedentary behavior, and lower overweight and obesity (20)(21)(22)(23). For example, one study showed that community residents tend to spend more time outdoors on weekdays and weekends, and also do more MVPA and vigorous activity, if they live in neighborhoods with more green spaces (24). Another study of Norwegian adults showed that living in a green community had a positive impact on increased PA among adults (25). According to a study of Danish adults, proximity to green space was associated with PA and obesity, and maintaining one's physical fitness was a key factor in choosing to visit green spaces (26). A study from American showed that BMI in adults was associated with forested land areas and proximity to public recreational areas (27). However, the current findings are inconsistent, such as a study from Portugal that did not find an association between the amount of urban green space, green exposure, and the prevalence of overweight and obesity in the Portuguese adolescent population (28). The effect of green space on PA and the obesity of residents still deserves further research.
China is rapidly urbanizing, modernizing, and growing economically. However, physical inactivity, sedentary behavior, being overweight, and obesity are prevalent among Chinese adults in the urbanizing and modernizing Chinese society (29,30). From 1991 to 2006, PA levels among Chinese adults declined at an unprecedented rate, with average weekly PA declining by 32%, and the largest percentage decline in PA in China compared to the United States, United Kingdom, India, and Brazil (31). In addition, since the 1990s, China has seen numerous significant social, economic, and lifestyle changes, which have coincided with the rapidly increasing prevalence of obesity. Over half of the adults and a fifth of children are overweight or obese, making China the nation with the highest proportion of such people globally (32). Consequently, the problem of physical inactivity and obesity among Chinese adults needs to be addressed.
Attempts to promote resident health behaviors through environmental interventions are receiving increasing attention, and green spaces may play an important role in environmental interventions. Despite the aforementioned work, two major gaps in the previous studies remain. First, research on green space and resident PA and obesity have focused on developed countries, with fewer studies targeting developing countries. For example, an American study showed that tree cover on streets was associated with active traffic behavior among residents and that a 10% increase in tree cover on streets was associated with a 19%-41% increase in the odds of active traffic (33). In addition, a study from England showed that people living in the greenest areas of England were more likely to achieve the recommended level of PA (34). The characteristics and quality of green space may vary from country to country due to differences in urbanization level, economic level, spatial scale, and geographic location, which may have different effects on the health behaviors of residents. For these reasons, research findings and recommendations for developed countries are not necessarily applicable to developing countries such as China. Therefore, in order to help establish a global framework for the use of green space, developing countries also need to actively contribute to scientific research.
Second, China is experiencing rapid urbanization and is facing serious physical inactivity and overweight/obesity problems, which according to existing studies are found to be regulated and controlled by environmental interventions. However, the results of current studies on the effects of green space on PA and the body weight status of Chinese residents are varied. The inconsistent results of previous studies challenge relevant policy makers and urban planners to develop targeted environmental intervention programs to improve the physical activity levels of residents. Therefore, a systematic review of existing studies is needed to better summarize the findings and limitations of existing studies and to make recommendations for future environmental interventions. In addition, the effects of green space on PA and the body weight status of residents may also vary by population, and no systematic review has been conducted specifically for Chinese adults.
Therefore, the purpose of this study was to systematically review the literature on the effects of green space on PA and body weight status in Chinese adults, to review the findings of existing studies, and to suggest future research trends. It also aimed to identify the research gaps in this field. The results of this review can provide a targeted theoretical basis for policymakers and stakeholders of environmental interventions.

Methods
This systematic review followed the recommendations provided by the Preferred Reporting Items for Systematic Reviews and Meta-Analysis guidelines (35
The exclusion criteria were as follows: (1) studies that did not include PA or body weight status as an outcome; (2) repeatedly published literature with poor quality assessment; (3) studies included Chinese subjects who resided outside of the Chinese Mainland, Hong Kong, or Macao; (4) case reports, review literature, conference papers, and dissertations; (5) non-English papers.

Study selection
The two researchers screened the literature independently according to the inclusion and exclusion criteria. First, the papers were initially screened by reading the title and abstract. Second, the full text was downloaded after obtaining the eligible literature, and full-text screening was performed. Finally, the two researchers compared the independently screened literature. For the literature with inconsistent screening results, the decision to include or not was to be made by a joint discussion with the third researcher.

Data extraction and synthesis
Information related to the included literature was extracted independently by two researchers using a standardized form. The extracted information was as follows: (1) basic information: first author, year of publication, and region; (2) basic characteristics of the participants: age, sample size, the proportion of females, sample characteristics; (3) statistical analysis: statistical model, nonresponse rate; (4) others: geographical coverage, type of green space measure, the detailed measure of green space, type of PA measure, detailed measure of PA, estimated effects of green space on PA, type of body weight status measure, the detailed measure of body weight status, estimated effects of green space on body weight status, and key findings. Because heterogeneous exposure and outcome metrics hindered Meta-analysis, we provided a narrative summary of common themes and findings from the included studies.

Study quality assessment
Each included study's quality was evaluated using the National Institutes of Health's Quality Assessment Tool for Observational Cohort and Cross-Sectional Studies (36). And each included study has 14 questions for the quality evaluation. For each question, "yes" responses received a score of one, while "no" responses received a score of zero. By computing the results for each criterion, a study's overall quality score is determined. The strength of the scientific evidence was measured using study quality assessment, however, the inclusion of research was not taken into consideration.

Search results
A total of 6,754 articles were retrieved from the database, and after removing duplicates, 6,099 articles were obtained. In the literature screening process, firstly, a total of 6,004 articles that did not match the research topic were removed by reading the titles and abstracts. Second, the remaining 95 articles were read in full text, among which 10 articles had environmental exposure indicators unrelated to green space, 5 articles had outcome variables that did not include physical activity and/weight status, 3 articles had study subjects that were not Chinese adults, and 6 other articles were excluded due to poor quality of literature and conference papers, and a total of 31 articles were finally included. Figure 1 shows the specific literature screening procedure.

Basic characteristics of the included studies
A total of 31 studies were included in this study, as shown in Table 2 (56,60,61) and three studies adopted experimental study design (49,53,55), respectively, and the remaining 25 studies used cross-sectional study design. The sample sizes of the included studies were generally large but varied widely among studies. Ten studies had sample sizes of less than 1,000, and the remaining studies had sample sizes between 1,000 and 40,000. All studies were of Chinese adults, aged ≥18 years, and 13 of the studies were of older adults aged 60 years or older. The proportion of females in each study ranged from 46.7% to 64.3%. A variety of statistical models were used in the studies, including multi-level logistic regression, structural equation model, multiple linear regressions, generalized linear mixed models, mixed-effect difference-indifference models, hierarchical linear model, order probit regression model, logistic regression, etc. The geographical coverage of green spaces is generally street greenery, parks, community green spaces, etc. The vast majority of studies were in urban areas, 4 studies were in urban and rural areas (42,47,51,61), and only one study was in a rural area (56). Table 3 summarizes the measures for green space, PA, and BMI in the included studies. For the measurement of green space, the majority of studies (n = 22) used objective assessment methods, four studies used subjective assessment methods, and five studies used a combination of subjective and objective assessment methods. The objective green space assessment methods mainly include geographical information systems (GIS), satellite remote sensing images, Baidu Street View map, Gaode Map, and Tencent Map using deep learning techniques. The assessment of subjective green space included self-reported questionnaires and field investigations. Eight studies examined the NDVI, eight studies examined the accessibility of green space, six studies examined certain characteristics of green, five studies examined the street greenness or green view index, and four studies examined the green space rate or green space perception.
For the measurement of PA, the majority of studies (n = 20) used a subjective assessment (e.g., self-reported questionnaires), only one study used an objective assessment (pedometer) (51), and one study used a combination of subjective and objective assessments (48). PA-related assessment indicators include the duration, frequency, and type of PA. For the measurement of BMI, 10 studies used an objective assessment and 3 studies used a subjective assessment (40,44,55). The body weight status measures included BMI, overweight, and obesity (e.g., peripheral obesity, central obesity).         Table 4 summarizes the effects of green space on PA and/or BMI in adults. Since different studies have used different green space measurement or evaluation methods, direct comparisons between studies cannot be made. This study starts from the conventional indicators of green space assessment, and divides the assessment of green space into green view index, green space ratio, accessibility of green space, and the design characteristics of green space. We summarize the key findings in the following four areas.

Key findings
First, all studies that used street-level green view index as a measure of green space showed that street-level green view index was significantly positively associated with PA and significantly negatively associated with BMI among Chinese adults. The green view index refers to the proportion of green scenes such as natural landscapes and vegetation in people's view (68). Street greenery, street greenness, and so on in the article represent the street-level green view index. For example, Studies have shown a positive relationship between street greenery and total walking time, with each standard deviation increase in street greenery being associated with an increase of approximately 0.2 standard deviations in walking time for older adults (64). Lu found that participants exposed to large amounts of street greenery were significantly more likely to engage in regular recreational greening activities compared to participants exposed to low amounts of street greenery (62). Moreover, research has revealed that residents' odds of being overweight or obese are higher in areas with a green view index of less than 15% (45). And with every interquartile range increase in street view greenness, BMI decreases by 0.15 kg/m 2 , the odds of being overweight decrease by 7%, and the odds of being obese decreased by 18% (52). Furthermore, in a natural experimental study, a greenway intervention was found to have a significant positive effect on walking time, especially for residents living within two kilometers of a greenway (49). Another recent study showed that greenway interventions were effective in reducing weight (55). However, in general, both longitudinal and experimental studies on the effects of green view index on PA and body weight status are scarce, and further research is needed in the future.
Second, most studies showed that green space ratio were associated with PA and/or weight status among Chinese adults, but some studies also showed no correlation. Green space ratio is the ratio of total green space to neighborhood area. The NDVI, green space rate, and so on in the article all represent the green space ratio. For example, compared to neighborhoods with a green space ratio of more than 28%, persons in neighborhoods with a green space ratio lower than 28% are more likely to be physically inactive (45). Wang et al. concluded that urban green spaces and open spaces play an important role in promoting PA among residents, especially women and the older adults, and their study found that the area of green space was significantly associated with residents' PA (39). However, the Huang et al. study found no connection between the NDVI and time spent on sedentary activities, leisure-related PA, or time spent traveling (42). The author explains that the research may focus on middle-aged and older adult people, who typically spend less time on traveling and leisure-related PA. NDVI may be difficult to affect the traveling and leisure-related PA of the older adults. In addition, half of the    1. Compared to participants exposed to low amounts of street greenery, those exposed to large amounts had a significantly higher likelihood of regularly engaging in recreational green PA (OR = 1.20, 95%CI = 1.08, 1.33).
2. Compared to residents who were exposed to low-quality street greenery, residents who were exposed to highquality street greenery were more likely to engage in frequent recreational green PA (OR = 1.10, 95%CI = 1.05, 1.25).
3. Compared to participants exposed to low levels of total park space in the buffer, those exposed to high levels of     9 Chen, 2020 When an urban green space was within a kilometer, it was linked to a lower risk of being overweight or obese (β = 0.320, p < 0.01).
Accessibility of green space related to obesity:      The density and accessibility aspects of parks can be a great stimulus to PA for older adults.
The density and accessibility of parks can reduce BMI in older adults.
The density and accessibility of parks are related to PA for older adults: +  Frontiers in Public Health 20 frontiersin.org participants live in rural areas where the vegetation is used for agricultural purposes and is not suitable for participating in sports activities related to transportation and leisure. It can be seen that individual differences, regional differences, and PA types of the research subjects may have different impacts on the results. Furthermore, in the reviewed studies, there was a significant negative correlation between the green space rate and obesity among Chinese adults. For example, in a longitudinal study design, scholars found that higher rates of green space reduced the risk of being overweight or obese (60). Third, the studies reviewed found a significant relationship between the accessibility of green space and PA and/or obesity among Chinese adults. The accessibility of green space in this study mainly refers to the convenience of reaching the green space from the residence, and the indicators for evaluation are mainly distance and time, etc. A previous study showed that the distance to a park was significantly and inversely related to the likelihood of residents to participate in PA. For every 1-unit (10%) increase in distance to a park, residents' participation in PA decreased by 18% or 27% (69). Wang et al. showed that the accessibility of green space was significantly and positively correlated with the PA of residents (39). A research study found that maximum park visits declined exponentially as travel distance to the park increased (46). Studies of older adults have found that being less than 800 m from the nearest park has a negative impact on sedentary activities, while a distance of 800-1,200 m may support those who regularly participate in recreational walking (50). In addition, easier access to urban green space was associated with lower odds of being overweight or obese when it was within 1 km of an urban green space (40). And a study by Xiao et al. reported that park accessibility was significantly and negatively associated with BMI in older adults (57). The accessibility of green spaces can reflect the ease of residents to reach green spaces such as parks, and also reflects the level of green space service capacity from the side. Based on the results of previous studies, it is recommended to improve the accessibility of green spaces around residences to promote residents' PA and reduce the risk of overweight and obesity.
Fourth, design characteristics of green space had an effect on PA and/or obesity among Chinese adults, but there were some studies showing no statistically significant results. The design characteristics of green spaces in this study mainly included variables such as outdoor fitness equipment in green spaces, the configuration of evergreen trees, the diversity of greenery, and residents' perceptions of green spaces. A study from Hong Kong showed that older adults' perceptions of park safety, attractiveness, and park characteristics were not significantly related to park-based exercise in Hong Kong's parks (65). And some scholars have also shown that space environment and landscape quality are not significantly correlated with residents' PA (39). However, studies have also shown that the total number of steps engaged by older adults is positively correlated with the presence of outdoor fitness equipment. And the energy expenditure of older adults was positively correlated with the presence of outdoor fitness equipment (48). Zhai and Baran showed that parkway length was positively associated with walking among older adults and that soft or level pavement, seating, flowers, and lighting fixtures were preferred when designing sidewalks (37). In addition, the research by Leng et al. shows that the design characteristics of green spaces (such as the evergreen tree configuration type in green spaces) are related to the overweight/obesity of residents (45). One longitudinal study showed that levels of residential greenery, tree cover, and tree-to-shrub/grass ratios were negatively associated with peripheral overweight/obesity and central obesity. Increased shrub, grass, and green diversity were associated with lower odds of peripheral overweight/obesity (61). The design characteristics of green spaces affect the perception and attractiveness of green spaces to residents. It is recommended that outdoor fitness equipment, increased diversity of greenery, and safety of green spaces be configured to increase the opportunities for residents to participate in PA. Table 5 lists the study quality assessment's global and criterionspecific scores. The average score for the included studies was 7 (ranging from 5 to 11). The research questions and objectives of all the studies included in the review were clearly stated, the study population was identified and defined with a 50% participation rate, participants were drawn from the same or related populations during the same period, and inclusion and exclusion criteria were pre-established and uniformly applied to all potential participants. Twelve studies investigated various exposure levels about the outcome. Twenty five studies used valid and reliable exposure measures. Twenty two studies implemented valid and reliable outcome measures. Twenty four studies measured and statistically adjusted for key potential confounding variables between exposure and outcome. In addition, six studies had fairly long follow-up periods, sufficient to observe changes in outcomes. And three studies assessed exposure more than once during the study period, and two studies had less than 20% loss to follow-up after baseline. In contrast, none of the studies provided a rationale for sample size using power analysis, none measured relevant exposures before outcomes, and none had outcome assessors blinded to the exposure status of participants.

Discussion
This study systematically reviews the scientific evidence on the effects of green space on PA and body weight status in Chinese adults. A total of 31 studies met the inclusion criteria for this study, including 25 cross-sectional studies, 3 longitudinal studies, and 3 experimental studies. Although green space was found to be significantly positively correlated with PA and negatively correlated with BMI, overweight, and obesity among Chinese residents in most studies, some studies showed no correlation. It is still difficult to draw a clear conclusion about the relationship between green space and PA and the body weight status of residents.
In this study, we found that street-level green view index were positively associated with residents' PA and negatively associated with BMI in studies focusing on street green space (52,58,64). Green space ratio in local spaces was positively correlated with PA for most of the studies reviewed, but negatively correlated with BMI for all studies (60). In studies focusing on the accessibility of green space, most studies showed that PA and body weight status of Chinese adults were significantly correlated with park accessibility (39,40). In addition, since the design of green spaces contains several dimensions, such as the diversity of green plants and the configuration of sports facilities       in green spaces. Therefore different design characteristics may have different effects on the physical activities of residents. For example, one study showed that natural space environment, landscape quality, and safety were not significantly associated with residents' PA (65). But one study reported that the presence of outdoor fitness equipment in parks was positively correlated with PA among older adults (48).
There are relatively few studies on green space characteristics and weight status of Chinese adults, which can be further explored in the future. In addition, a total of three studies in this systematic review used an experimental design, and all showed positive effects of green space interventions on residents' PA and weight status (49,53,55), but there are still relatively few studies using an experimental design, and further relevant studies are needed to support this result in the future.
In general, the main findings of the systematic review are similar to those of studies in a number of developed and developing countries. For example, in developed countries, a previous study explored the relationship between accelerometer-based measures of PA levels of older adults in Spain and their visits to green spaces. It showed that visiting green spaces was positively related to older adults' PA levels (70). And in a study of Japanese older adults, the presence of parks or green spaces was found to be positively correlated with the frequency of PA (71). Another study from the Netherlands showed that the odds of children being overweight decreased with increasing exposure to green space in a 3,000 m buffer zone (72). Moreover, among developing countries, a study from Turkey showed that short distances to urban green spaces, many trees, exercise equipment and picnic areas were positively associated with the frequency of physical activity of residents (73). And in one Dutch city, the amount of green and recreational space, especially parks and sports fields, within 300 and 500 m of participants' homes was positively correlated with time spent cycling (74). Studies of Mexican adults showed that exposure to green space was associated with lower BMI, even after adjusting for important confounders (75). In our systematic review, most studies also showed that green space had a positive effect on PA and weight status of Chinese residents. This further confirms that environmental interventions can be used to promote healthy behaviors and increase physical activity levels among residents.
However, contrary findings also exist, for example, a study from New Zealand showed no significant correlation between neighborhood green space and overweight and obesity in children (76). In addition, a cross-sectional study of 3,679 Scottish adults showed no significant correlation between the availability of green space in the community and the total PA of residents (77). There were also studies in the literature included in this study that did not find a relationship between green space and PA or weight status. The potential explanation for this inconsistency between studies could be due to differences in the way green space was measured or the indicators were chosen to assess it between studies. The green space data sources of the included studies differed, such as from satellite remote sensing images, self-reported questionnaires, individual street maps, and regional government databases, and thus would provide varying degrees of accuracy. For example, studies have shown that street-level green views assessed using Google Street View images show a significant correlation with the odds of residents bicycling, while overhead green views assessed by NDVI are not correlated with the odds of bicycling (78). In addition, confounding may be a potential factor for differences in results between the studies included in this systematic review. Most studies included only socio-demographic variables for adjustment, such as gender, age, Criterion Study ID 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29   Frontiers in Public Health 25 frontiersin.org education level, and economic level, and few studies included variables such as social cohesion, air quality, and noise for adjustment. This study systematically and comprehensively reviewed the existing literature on the effects of green space on PA and body weight status of Chinese adults. It is recommended that future studies should research environmental effects on human health behaviors for different green space characteristics, exposure doses, and different populations to better understand the relationship between the environment and human health. Second, most of the current studies mainly use cross-sectional study designs, which cannot explain the causal relationship between green space and health, and more longitudinal studies or experimental studies need to be used in the future to further explore the effects of green space on residents' health. Third, the dose and duration of green space exposure affecting PA and weight status of Chinese residents are still not fully understood, and the dose-response of green space exposure and the duration of its effects on residents' health needs to be analyzed in the future. Finally, further analysis of the mediating factors affecting the relationship between green space and residents' health, such as socioenvironmental factors, air quality, and noise, is needed in the future, and these complex causal relationships between environmental characteristics need to be further elucidated in future studies.
The results of the systematic review provide some guiding information for the development of relevant sectoral policies and environmental intervention programs. However, there was a large heterogeneity of green space exposure and study participants in the included studies; and there was a large variation in the assessment of green space and PA across studies. For example, for the assessment of green space, some studies assessed the scale of green space while others assessed the perception of green space; for the assessment of PA, accelerometry and self-reported questionnaires were included. In addition, different studies control for different confounding factors, which can produce some heterogeneity. The above reasons prevented us from conducting a quantitative and comprehensive analysis. Second, not all of the included studies used objective assessments of green space and PA measures, and some of the studies used self-report questionnaires, so the questionnaire results may have produced some recall biases. Third, only published literature in English was included in this study, excluding literature in other languages, and there may be some selection biases.

Conclusion
This systematic review provides a comprehensive overview of the effects of green space on PA and body weight status among Chinese adults. Green view index, green space accessibility, and green space ratio were found to be positively associated with PA and negatively associated with BMI. Furthermore, green space interventions were effective in increasing PA levels and managing healthy weight among Chinese adults. However, the relationship between the design characteristics of green space and PA and weight status still needs to be further explored. It is recommended that more longitudinal and quasi-experimental studies be conducted in the future to explore the effects of green space on PA and the body weight status of residents and the causal relationships among them, to provide some evidence to support relevant policy formulation and intervention program design.

Author contributions
YS: study concept and design and drafting of the manuscript. HL: data extraction and analysis. HY: study supervision. All authors contributed to the article and approved the submitted version.

Funding
This study was supported by the National Social Science Foundation of China (17CTY020, 20BTY004), Beijing Social Science Foundation of China (21YTA009), and the Tsinghua University "Shuang Gao" Scientific Research Program (2021TSG08208), the Tsinghua Education Reform Project (2021ZY01_01), the Tsinghua Graduated Education Reform Project (202303J039).